JP3071085B2 - Chromatogram analysis display method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a chromatograph and, more particularly, to a method and an apparatus for displaying a chromatogram analysis result.

[0002]

2. Description of the Related Art As a method of determining the reliability of a qualitative analysis of a chromatogram, there has been proposed a "chromatographic spectral data processing method" described in JP-A-4-212059. In this method, Student's t-test, which is a statistical method, is performed based on UV spectrum data from a multi-channel detector, and the reliability of peak identification is calculated. In addition to the spectral data, the chromatogram features such as retention time, peak area, and peak height are also analyzed. The smaller the deviation from the feature of the standard sample, the more the sample score approaches 0. Defined and the reliability of peak identification is determined.

Similarly, a method of diagnosing the reliability of peak identification based on the ratio between the peak area and the peak height has been proposed in Japanese Patent Publication No. Sho 61 (1986).
No. 54181. Here, a calculation formula used for diagnosis is described. It also touches on reliability indicators and how to display them.

[0004] Further, in a chromatographic data processing apparatus described in Japanese Patent Application Laid-Open No. 63-151851, a method of decomposing into two peaks for detecting overlapping peaks is described. That is, one of the obtained chromatograms is multiplied by the coefficient α by the two different detection wavelengths, and this is subtracted from the other chromatogram. Further, one chromatogram is multiplied by the other coefficient α ′, which is subtracted from the other chromatogram. Then, from these results, the position and width of the first peak and the position and width of the second peak are calculated. The calculated first and second peaks are set as initial values, and complete peak analysis is performed by the nonlinear least squares method.

[0005]

By the way, if an operator using a chromatographic apparatus can recognize a quantitative value such as concentration and the reliability of peak identification, not only the reliability of the data processing method, The overall reliability of the entire chromatogram analysis can also be grasped quantitatively.

However, the spectrum data processing method described in Japanese Patent Application Laid-Open No. 4-212059 is a method of calculating the reliability of qualitative analysis, and does not mention the reliability of quantitative analysis such as concentration (kg / m ³ ). Not.

On the other hand, as for the reliability related to the qualitative analysis, first, an expensive multi-channel detector is required for spectrum matching. In addition, the peak groups having similar retention times often have similar molecular structures, and the difference may not appear in spectral matching.

Further, in the above spectrum data processing method, a sample score is defined. However, even if the expression is statistically accurate, it is not understood by a general operator as a reliability. This is a difficult value. For the reliability of the qualitative analysis of chromatograms, an expression method that is easier to understand is desired.

Further, in the qualitative analysis of chromatograms, that is, in peak identification, the most basic information is the retention time, but the reliability (error) of this time is not considered in the above data processing method. Not discussed. If the retention time that is the basis for peak identification is uncertain, the reliability of peak identification itself will be low. Therefore, in the qualitative analysis of the chromatogram, it is desired to calculate the reliability of the retention time and calculate the reliability of the qualitative analysis based on the calculated reliability.

In particular, attention must be paid to the reliability of the retention time when two or more peaks overlap. Even if the peaks clearly appear at the peaks, they interact with each other, and the peaks shift from the true retention time. If the time at the top is recognized as the retention time as it is, a peak may not be correctly identified because it contains a large error.

Further, when one peak is significantly smaller than the other peak, when they overlap, a so-called shoulder peak is formed. At this time, the retention time of a particularly small peak is considerably uncertain. In the data processing, the inflection point is regarded as the retention time, or as described in JP-A-63-151851, a non-linear least squares method is used to decompose the overlapping peaks into a data processing time. Can be requested. Regardless of the method used, the retention time of the shoulder peak is generally found to include a considerable error.

For this reason, it can be easily considered that the shoulder peak has a higher probability of erroneous peak identification than a clear peak. From such a viewpoint, it is understood that it is important to calculate the retention time error and calculate the reliability of peak identification.

In the method of diagnosing reliability of peak identification described in Japanese Patent Publication No. 61-54181, there is no description about the reliability of quantitative analysis or the reliability of retention time in peak identification. The reliability index of peak identification is also for convenience and is not easily understood.

An object of the present invention is to determine a quantitative value such as a concentration and a time such as a retention time with high accuracy, obtain an error with respect to the obtained numerical value, and calculate a quantitative value and a reliability of peak identification. An object of the present invention is to realize a chromatogram analysis display method and a display device capable of displaying.

[0015]

The present invention is configured as follows to achieve the above object.

[0016] In a chromatogram analysis and display method for analyzing chromatogram data and displaying an analysis result, a step of detecting a peak of the chromatogram data, a step of identifying a peak of the chromatogram data, and a step of detecting and identifying the peak detected and identified. Based on peak size corresponding to
Te, calculating the analysis value of the object to be measured, the error of the calculated analysis values, and calculating by using the peak size, the calculated analytical values, a display unit and an error of analysis Displaying.

Also, in a chromatogram analysis and display device for analyzing chromatogram data and displaying the analysis result, a peak detection unit for detecting a peak of the chromatogram data, a peak identification unit for identifying a peak of the chromatogram data, based on the peak size corresponding to the detected and identified peaks, and the analysis <br/> value calculation unit for calculating an analysis value of the object to be measured, by using the peak size was calculated min
It comprises an error calculator for calculating an error of analysis values, and the calculated analysis values, and a display unit for displaying the error in the analysis.

Preferably, in the above-described method and apparatus for analyzing and displaying chromatograms, the error of the analysis value is calculated including the reliability of the baseline of the chromatogram data. Preferably, in the method and apparatus for analyzing and displaying chromatograms, the error in the analysis value is calculated including the reliability of dividing peaks adjacent to each other.

[0019]

Further, preferably, further comprises in the chromatogram analysis display method, a step of calculating an error of peak size, and a step of displaying the error in the calculated peak size on the display means.

Preferably, in the above-mentioned chromatogram analysis display device, the error calculation unit calculates a peak size error, and the display unit further displays the calculated peak size error. Preferably, in the method and the apparatus for analyzing and displaying a chromatogram, the peak size is a peak height.

Preferably, in the method and apparatus for analyzing and displaying a chromatogram, the peak size is a peak area. Preferably, in the method and apparatus for analyzing and displaying a chromatogram, the peak size is a peak height and a peak area.

[0023] In a chromatogram analysis and display method for analyzing chromatogram data and displaying the analysis result, a step of detecting a peak of the chromatogram data, a step of identifying a peak of the chromatogram data, Calculating the calculated peak retention time, calculating the error in the calculated peak retention time, and displaying the calculated peak retention time and the error in the retention time on the display means. Prepare.

Also, in a chromatogram analysis and display device for analyzing chromatogram data and displaying the analysis result, a peak detection section for detecting a peak of the chromatogram data, a peak identification section for identifying a peak of the chromatogram data, A retention time calculation unit that calculates the retention time of the detected and identified peak, an error calculation unit that calculates an error of the calculated peak retention time, and a calculated peak retention time, and an error of the retention time. A display unit for displaying.

Preferably, in the method and apparatus for analyzing and displaying chromatogram, the peak is an inflection point of the chromatogram data. Further, preferably, in the chromatogram analysis display method, a step of calculating the reliability of the peak identification based on the error of the calculated peak retention time and the peak retention time, and calculating the calculated peak identification reliability. Displaying on the display means.

Preferably, in the above-mentioned chromatogram analysis display device, the error calculating section calculates the reliability of peak identification based on the calculated peak holding time and the error of the peak holding time. The calculated reliability of peak identification is further displayed.

Preferably, in the method and apparatus for analyzing and displaying chromatograms, the error of the calculated peak retention time is calculated including the reliability of dividing peaks adjacent to each other.

[0028]

[0029]

Further, in the chromatogram analysis and display method for analyzing the chromatogram data and displaying the analysis result, the step of identifying a maximum point of the chromatogram data within a predetermined range includes the steps of: Calculating a search section that does not include the maximum point; calculating a polynomial curve that fits the chromatogram data in the calculated search section; detecting an inflection point of the polynomial curve; A peak calculation step for calculating an error of the peak holding time corresponding to the detected maximum point and the inflection point, an error of the peak holding time, a peak height, a peak area, a peak area percentage, and an error of the peak area percentage, and the calculated peak The display means displays the retention time, the peak height, the peak area, the peak area percentage, and the error of the peak area percentage on the display means. Tsu and a flop.

Preferably, in the above-mentioned chromatogram analysis display method, the peak calculation step calculates a peak height error and a peak area error, and the display step includes a peak holding time, a peak holding time error, The display means displays the peak height, the peak height error, the peak area, the peak area error, the peak area percentage, and the peak area percentage error.

Preferably, in the method for analyzing and displaying a chromatogram, a step of calculating a search section;
Determining a polynomial curve between the step of determining whether there is a local maximum point in the calculated search section,
Setting a search section obtained by expanding the search section by a predetermined width when there is no local maximum point in the search section.

Preferably, in the method for analyzing and displaying chromatograms, the likelihood of the identified and detected maximum point and inflection point is calculated using an index that is a probability, a difference, a confidence interval or a confidence rate; Displaying the calculated index on the display means.

Further, in the chromatogram analysis display device for analyzing the chromatogram data and displaying the analysis result, a detecting means for detecting a component contained in the sample and a storage unit for storing the chromatogram data detected by the detecting means are provided. A peak identification unit that identifies a maximum point of data within a predetermined range of the chromatogram data stored in the storage unit, and a section calculation that calculates a search section that is near the identified maximum point and does not include the maximum point. Unit, a fitting unit that calculates a polynomial curve that fits the calculated chromatogram data in the search section, and a differential feature point search unit that calculates the differential value of the polynomial curve and detects an inflection point from the calculated differential value And the peak holding time and peak holding time error and peak height and peak area and peak area percentage corresponding to the identified maximum point and inflection point, and Comprises an error probability calculation unit for calculating the error of the over-click area percentage, the calculated the peak retention time and peak height and peak area and peak area percentage and the peak area percentage and a display unit for displaying an error.

Preferably, in the above-mentioned chromatogram analysis display device, the error probability calculation unit calculates an error of the peak height and an error of the peak area, and the display unit displays the error of the peak holding time, the error of the peak holding time and the peak height. The peak height error, the peak area, the peak area error, the peak area percentage, and the peak area percentage error are displayed.

Preferably, in the above-mentioned chromatogram analysis and display device, the error probability calculation unit calculates the likelihood of the identified and detected local maximum point and inflection point as an index which is a probability, a difference, a confidence interval or a confidence rate. And the display unit displays the calculated index.

Preferably, in the above-mentioned chromatogram analysis display device, the error probability calculating section changes the error of the peak area according to the shape of the baseline for determining the peak area.

[0038]

[0039]

When calculating the error of the peak size such as the peak area and the peak height, it is general to use a noise value. If two or more peaks overlap, the error may vary depending on the method of dividing the peak area. In the case of performing peak decomposition in data processing using the non-linear least squares method, the peak size is determined relatively accurately, and the error is estimated to be small. Also, if the baseline is not simply horizontal,
The accuracy of the peak size deteriorates, and the error is largely estimated.

The quantitative value is generally determined from the peak size of the measurement sample with reference to the peak size of the standard sample. Therefore, the error of the quantitative value is propagated by the error of both peak sizes. In consideration of this, the error of the quantitative value is calculated. Repeated measurement of a standard sample or accurate determination of a calibration curve using a standard sample having a different concentration is nothing less than minimizing the error propagated from the standard sample after all.

In the calculation of the retention time error, the retention time error is estimated using the noise value at the clear peak of the vertex. There is also a method of estimating the time error of the maximum point by the smoothing process.

If two or more peaks overlap, data processing is performed by a non-linear least squares method or the like, the peaks are decomposed, and the time error of the vertices of each decomposed peak is statistically calculated. .

In the case of a shoulder peak, the inflection point may be regarded as the retention time for convenience, and the time error of the inflection point may be statistically calculated. Further, the reliability of peak identification is calculated from the calculated retention time error as follows.

The reliability of peak identification is generally obtained from the retention time of a standard sample and its error and the retention time of a measurement sample and its error. From these, a section where a true difference δ is likely is estimated. The reliability of peak identification is calculated based on the obtained estimated section. For example, a reliability rate (%) defined for convenience is obtained.

By outputting the error of the quantitative value, the reliability of the obtained measurement result can be tested. For example, even if the result is simply 10.0 ppm as in the past, is the true value between 9.9 ppm and 10.1 ppm, high accuracy, or between 5 ppm and 15 ppm, low accuracy? Was difficult to determine.

In the present invention, for example, 10.0 ± 0.0.
Since 1 ppm is displayed including the error, the reliability of the measurement result can be recognized. Similarly, the reliability of peak identification can be recognized based on the retention time error.

[0047]

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a flowchart showing the operation of the chromatogram analysis and display method according to one embodiment of the present invention.
The embodiment of FIG. 1 is an example applied to the analysis of glycohemoglobin by ion exchange chromatography. FIG. 3 is a schematic configuration diagram of a liquid chromatograph device for analyzing glycated hemoglobin by the ion exchange chromatograph.

First, in FIG. 3, the eluent liquid sending pump 40 is operated by the command signal of the control unit 44 so that the eluent A 48, the eluent B 49, and the eluent C
50 are switched at a cycle of 1.9 minutes, 1.0 minutes, and 0.4 minutes, respectively, at 3.3 minutes, and are sent to the separation column 41. A sampler (sampling means) 43 having a movable needle aspirates 5 μl of a standard sample 52 or an unknown sample 51 of hemoglobin (Hb), dilutes it by hemolysis, and sends it to the separation column 41 together with the eluent A48. Then, the components contained in the sample are separated and developed, and are detected by the visible absorbance detector 42. Then, the chromatogram, which is the detected data, is supplied to the data analysis unit 45 and stored.

FIG. 2 is a display example displayed by the CRT 46 or the printer 47. In this display example, the waveform displayed on the display unit D3 is a chromatogram. In this chromatogram, component LA-1c and component SA
The separation from component 1c is insufficient, and the component LA-1c contains a shoulder peak. If the maximum point is regarded as a peak, the shoulder peak of this component L-A1c cannot be identified as a peak. The shoulder peak is recognized as an inflection point, that is, a point at which the second derivative becomes zero. Hereinafter, an inflection point is obtained by the data analysis unit according to the following procedure.

In step 1 of FIG. 1, the operation flow is started when the digitized chromatogram is obtained. Next, in step 2, the maximum point in the time window is identified as the peak of the component S-A1c. Here, the time coordinate of the maximum point is recognized as the retention time of the component S-A1c. In step 3, the component L-A1
The search section for the peak c is calculated. This search interval is based on the obtained retention time t _s-A1c of the component S-A1c, for example, [t _S-A1c −0.45, t _S-A1c −
0.15]. This means that the center of the time window is t _S-A1c −0.30 (min) and its width ± 0.1
This means that the peak of the component L-A1c is identified as 5 (min). This search section can be complicated as at _S-A1c + b using a and b as coefficients.

Next, at step 4, it is determined whether or not there is a maximum point. That is, the presence or absence of a local maximum point in the search section is checked. If there is no maximum point, the process proceeds to step 5.

In step 5, in order to search for a shoulder peak, a section in which a chromatogram data point is fitted to a cubic curve is calculated and set. The reason for the cubic curve is that it is a minimum-order polynomial having an inflection point. Generally, it is sufficient if the curve is a third-order or higher curve. The fit interval is slightly wider than the peak search interval of the component L-A1c,
[T _S-A1c -0.55, t _S-A1c -0.05] are set. Alternatively, the section can be set using the time t _{1 of the} peak start point.

Next, in step 6, the chromatogram data points in the above-mentioned fit section are converted into a cubic polynomial y ₃ = a
Fit x ³ + bx ² + cx + d using the least squares method. Then, in step 7, the inflection point of the polynomial curve is searched to find the shoulder peak, and the following judgment is made. (1) Did it fit well? (2) Is there an inflection point clearly? The fitness of the fit in the determination (1) is determined based on, for example, whether the least square value is equal to or less than a specified value. The clarity of the inflection point in the judgment (2) is, for example, simultaneously fitted to the linear expression y ₁ = ex + f, and if the area surrounded by the cubic expression y ₃ and the linear expression y ₁ in the fitting section is equal to or larger than a specified value. It is determined by whether or not there is. Or, simply the value of the cubic equation y ₃ of the third-order coefficient a can be determined by whether or not more than a specified value. If either of the determinations (1) and (2) is not satisfied, the component L-A1c is recognized as not being present as a shoulder peak, the process proceeds to step 16, and the process ends.

If it is determined in step 7 that both the determinations (1) and (2) are satisfied, the process proceeds to step 8.
In step 8, the fitting section is set again based on the coefficient obtained in step 6, and the fitting is performed more precisely. For example, based on the inflection point time t _i = −b / 3a, the refit section is set to [t _i −0.15, t _i +0.1
5]. Similar to step 6 above, fitting is performed again to the cubic expression, and coefficients a, b, c, and d are obtained again.

Next, the routine proceeds to step 9, where the inflection point coordinates are calculated. In this step 9, the inflection point time t
Calculate _i as t _i = −b / 3a. The error e _ti of time t _i is, a and b, respectively e _a fit error of as e _b, is obtained by the following equation (1). e _ti = t _i √ ((e _a / a) ² + (e _b / b) ² ) (1) where t _i = −b / 3a.

Then, in step 10, the area occupied by each component and its error are calculated based on the time t _i and error e _ti obtained above. Next, in step 11, the percentage of the area occupied by each component and its error are calculated. In step 13, the calculated percentage of the area and its error are displayed as shown in FIG. That is, the percentage of the area occupied by each component and its error are displayed near the center of the display example shown in FIG.
For example, the component A1a is displayed as 0.9 ± 0.0%, and the component A0 is displayed as 93.3 ± 4.7%.

Subsequently, in step 14, the retention time for each component and its error are displayed in the display example shown in FIG. For example, in the case of the component A1a, 0.24 ± 0.01% on the right side of the percentage and the error thereof,
In the case of the component A0, 2.48 ± 0.02% is displayed. Next, in step 15, the calculated area of each component and its error are displayed in the display example shown in FIG.
For example, in the case of the component A1a, 13969 ± 698, the component A0
If so, it is displayed as 1382244 ± 69112.

As will be described later, when the calculated peak probability is 80% to 50%, a warning mark * is displayed at a portion indicating the component. When the calculated probability of the peak is less than 50%, the part indicating the component has a double mark? Is displayed.

If the maximum point of the component L-A1c is identified in step 4, the process proceeds to step 12, where an error is calculated, and the process proceeds to step 10. Error calculation of the step 12 calculates an error e _tm maximum point time t _m. Briefly, as shown in FIG. 4A, a threshold value of (maximum value) −2 × (noise value) is set using a noise value acquired in advance, and the left and right of the maximum point time t _L-A1c are respectively set. , t _-, as t _+, the better the time is further away from the t _m and the limits of error. That is, the larger of | t ₋ −t _m | and | t ₊ −t _m | is recognized as the error _etm . Further, FIG.
As shown in, after once smoothing the vicinity of the local maximum point, it is also possible to estimate the error in t _m corresponding to the noise value. The obtained t _m and e _tm are output in step 14 as t _m ± e _tm .

In general, the components LA-Ic and SA-I1c are quantitatively determined by drawing a perpendicular line from the valley if there is a valley between peaks and from the inflection point if there is a valley between the peaks and dividing the area. Component LA
The ratio of each of 1c and component S-A1c to the total peak area is indicated by percentage.

In FIG. 2, the display area D1 shows the percentage of the area of the components L-A1c and S-A1c and its error, the percentage of the total area of the components L-A1c and S-A1c and its error, and the component. A1a, A1b, L-A1c
And the percentage of the total area of S-A1c and its error are displayed.

Further, as described above, the display unit D2
The area percentage for each component and its error, the retention time and its error, the area and its error, a warning mark, a doubt mark, and the total area of the area of each component are displayed. In addition, as described above, the name of each component, the percentage of the area and its error, a warning mark, and a doubt mark are displayed on the display unit D3, together with the chromatogram.

FIG. 5 shows another display example different from the display example shown in FIG. In the example of FIG. 5, the concentration (mg / l) of each component and its error are displayed, but no error is displayed for the retention time and the area. The other parts are the same as the example of FIG. 5 and the example of FIG. It is also possible to display as in the example of FIG.

Next, another example of the inflection point search will be described. For example, in amino acid analysis, there is a characteristic baseline fluctuation that rises near NH3 as shown in FIG. In this chromatogram processing, similarly to the above-described embodiment, the inflection point search section is set to, for example, [t ₁ +0.10, t _NH3 −0.10].
Set as Here, t ₁ is the time of the rising point, and t _NH3 is N
This is the retention time of H3. Hereinafter, in the same manner as in the example of FIG. 1, the fitting of the cubic curve is executed, the inflection point is obtained, and the inflection point is recognized as the start point of the baseline.

Now, an error calculation method will be organized and described. 1. The time error of the inflection point is 3 as shown in the above equation (1).
The calculation is performed using the coefficient at the time of fitting to the next curve and the error of the coefficient. This is used as the retention time of the shoulder peak or the time error between the start and end points of the baseline.

2. As for the time error of the local maximum point, a time change corresponding to noise is calculated as shown in FIG. Or 2
It can also be calculated from the coefficients by fitting with a quadratic curve or the like. This is used as an error of the retention time of a peak having a normal term point.

3. The peak area error e _A can be calculated, for example, as in the following equation (2) (Grus, Reference)
hka, E .; Zamir, I. "High Performance Liquid Chroma
tography "; John Wiley & Sons: New York, 1989; Chapt
er 13.). _{e A = A√ (2 / π} ) · (N / H) (1 / n) --- (2) However, A is the peak area, H is the peak height, N is the noise value, n represents the number of sampling points It is.

Since the largest factor for peak area variation is the way of drawing the baseline, an algorithm that estimates the error larger for a chromatogram with a higher dependency on the baseline can also be used. (A) of FIG.
FIG. 7B is an example of a chromatogram in which a peak area is obtained without depending on a baseline, and FIG. 7B is an example of a chromatogram depending on how to draw a baseline.
Further, as shown in FIGS. 7C and 7D, the peak area varies depending on the method of dividing the overlapping peaks. Generally, the calculation as in the above equation (2) is sufficient.

4. The error of the peak height is (noise value) √
Estimate about twice. In this case as well as the peak area, the dependence on the baseline can be considered.

5. Error e of quantitative value such as concentration of unknown sample
_Q propagates an error in a peak area or a peak height.
When the proportionality constant for quantitative calculation is determined from the peak area of the standard sample or the like, it is determined by the following equation (3).

E _Q = Q√ ((e _S / S) ² + (e _U / U) ² ) (3) where Q is a quantitative value, S is the peak area of a standard sample, and e _S
Is the error of S, U is the peak area of the unknown sample, and e _U is the error of U.

In the case of quantification using a calibration curve, similarly, error propagation from the calibration curve can be calculated.

Up to this point, the error in the data processing has been estimated. However, when other statistical errors e _sta are taken into account, the data is input from the input unit 54 and the CRT 46 or the printer is input in accordance with the following equation (4). Output to 47.

[0074] ^{_{e out = √ (e d 2}} + e sta 2) --- (4) However, e _out is output to the error, e _d is the error calculated by the data processing.

Further, when a systematic error such as correction is added, an input operation is required. To calculate such an error, a mathematical expression can also be input from the input unit 54. Also,
Various errors corresponding to each component are stored in a table,
It can be read at the time of calculation.

The error can also be used when calculating the membership function of fuzzy inference.

Next, an analysis value obtained in the chromatogram analysis of the retention time and its error, etc., is input, an identification probability value as to what component an arbitrary peak is is output, and the input and output are fuzzy inferred. An example of a qualitative analysis method that is linked by the following will be described.

As shown in FIG. 8A, the proposition "1.0
The peak appearing at 0 min is F. Is a proposition corresponding to the time window. For example, assuming that the time window center and width of the membership function W _i (t) are 1.00 min and 0.15 min, respectively, the function W
_i (t) represents the center of gravity of 1.00 m as shown in FIG.
in, ± 0.15 min for the upper base and ± 0.3 for double the lower base
It is set to a trapezoid of 0 min. On the other hand, a proposition "a measurement peak appeared at about 0.9 min." As shown in FIG. 8B, the membership function r _j (t) sets the time and the error of the local maximum point to 0.90 min, respectively.
If it is 0.20 min, the normal distribution is set with a vertex of 0.90 min and a standard deviation of 0.20 min.

The probability that the measured peak is F is determined by the two membership functions W as shown in FIG.
_i (t) and r _j (t) are overlapped to determine an overlapping portion. In brief, the maximum point of the overlapping portion is defined as the coincidence, and the coincidence is interpreted as the probability of being F.

If this probability becomes 80% or less, the output result of the CRT 46 or the printer 47 is shown in FIG.
A warning mark * is displayed in the display example of. further,
When the probability becomes 50% or less, doubly mark? Is displayed.

As a stochastic / statistical expression, it is possible to display "5% risk rate, peak 1 is component A1a." In this case, assuming a normal distribution,
It is common to calculate the risk factor.

Further, when the probability is to be determined precisely, the components A1a, A1b, F, and L are set as shown in FIG.
-Set membership functions corresponding to the time windows of A1c, S-A1c, and A0. Of function w _{i i} =
.., 6 correspond to A1a to A0, respectively. The function w ₇ corresponds to the time window of the impure component,
As shown in FIG. 9B, the function is made to complement the function group shown in FIG. 9A. These functions w _i are:
Although it is trapezoidal, a normal distribution shape may be used. Operation
The user can input the membership function in an appropriate form.

Now, the membership functions in consideration of the error in the retention time of the measured peak are obtained as shown in FIG. 9 (A). To determine the probability, the degree of coincidence C _ij is calculated as in the following equation (5), and the probability P _ij is determined by the following equation (6).

[0084]

(Equation 1)

Here, P _ij is the probability that the j peak is the i component (i = 1 (A1a), 2 (A1b), 3 (F), 4
(L-A1c), 5 (S-A1c), 6 (A0), 7
(Impurity component)), n is the number of components.

For example, in the chromatogram shown in FIG.
The probability that the (j = 3) th peak is F is determined. At the same time, the probability of the impure component IMPU is determined. There is also a method of displaying the probability as a percentage in a pie chart or the like, but as shown in FIG. 10, the component name of the first candidate (1ST) and the probability (PRO
B), component name and probability of the second candidate (2SD) (PRO
B), quantitative values (%) (QUAN (%)) can be output as a table. The second candidate is output when the identification probability of the first candidate is 0.90 or less.

Further, in order to improve the reliability of identification,
Two components are arbitrarily selected from among the detected peaks, and it is confirmed whether or not they have a previously derived correlation. For example, when the peak 2 and 3 have been identified as A1b and F components, retention time t ₃ of the retention time peak 2 t ₂ and peak 3, t _A1b
It is checked whether the relation of 0.90 t _F -0.30 (min) is almost satisfied. Here, t _A1b and t _F are A1
This is the retention time of b and F. Here, in the case of six components, 15 combinations of ₆ C ₂ = 15 correlations are confirmed. And if there is a peak that deviates from the specified value, fuzzy inference is performed again,
Obtain the optimal solution. The area of each membership function W _i (t) or r _j (t) can be normalized to 1 as necessary.

Another statistical method for estimating the probability of peak identification will now be described. Basically, peak identification compares the retention time of a known peak in a standard sample with the retention time of an unknown peak in an unknown sample,
The one with the small difference is to correspond. It is considered that the smaller the accuracy (accuracy and precision) of this difference, the lower the risk of causing an identification error, that is, the higher the reliability of identification. This can be evaluated using statistical techniques (Reference, Okuno, Haga, Experimental Design, Baifukan, 1969).

First, the processing is started from step 70 in FIG. Then, in step 71, the retention time t _s of the S-A1c peak in the standard sample and its error σ _s are estimated. This may be the method shown in FIG. 4 described above, or uses the _first and _second moments μ ₁ and μ ₂ as shown in the following equations (7-1), (7-2), and (7-3). You can also.

[0090]

(Equation 2)

Μ ₀ , μ ₁ , and μ ₂ correspond to the peak area, the center of gravity, and the variance (variance), respectively, as shown in FIG. 12 (refer to “Theory a” by Abdel Salam Said, reference).
nd Mathematics of Chromatography ", Huthing, Hei
delberg, 1989). μ ₁ is the retention time, √
(Μ ₂ / N) is the retention time error. Here, N is the number of sampling points actually integrated.

Further, practically, based on the method of moment, the half width of the peak or 1 / √ (e) of the peak height is used.
Gaussian standard deviation σ from width at 0.607 times
And an error in the retention time can be estimated from σ / √ (N). Here, N is the number of data points forming a peak. The retention time may be a first moment or an average value of an appropriate section, but is suitably obtained from the local maximum using the method shown in FIG. This practical method is characterized in that the retention time and its error can be estimated with relatively little interference even for overlapping peaks.

Although the above description has been made on the analysis of one peak, the standard sample may be repeatedly measured a plurality of times, and a statistical error (accidental error) may be adopted.

Next, in step 72, the retention time t _u of the peak 5 in the unknown sample and its error σ _u are obtained by the above-mentioned practical method. Next, in step 73, t _S, t _u, using sigma _s, sigma _u, determine the true difference δ is likely interval between t _s and t _u for the reliability of 95%. The narrower this section is, and the closer it is to 0,
It can be considered that the reliability of identification is high.

First, the measured difference d is d = t _u −t _s . This difference d can be displayed as shown in FIG. Also, the estimated value s _d of the standard deviation of the difference _d
Is determined by s _d = √ (σ _s ² + σ _u ² ). Degree of freedom f
Is 2N-2. Here, N is the number of data points forming a peak. Even if N is infinity, there is no problem in performing the t test.

[0096] Here, if, when adopting the random error due to above repeated measurements of the sigma _s, considering the degree of freedom of sigma _s and sigma _u, it is necessary to weighted average. At this time, since the calculation is complicated, for example, the equivalent degree of freedom must be calculated by the Sata's weight method, σ _s
It is practical to use the method of substituting σ _u or σ _u . (R-
1) Compare rσ _S ² and (N−1) Nσ _u ² , and consider that the larger one is dominant in the error variance, and _let s _d be the larger σ _s or σ _u and ignore the other Resulting in. The degree of freedom f used is the degree of freedom r-1 or N-1 on the adopted side. Here, r and N are the number of repeated measurements of the standard sample and the number of data points forming a peak, respectively.

Next, since the standard deviation s _d and the degree of freedom f of the differences d and d have been obtained, in step 74, an estimated section of the true difference δ is calculated by the following equation (8). δ = d ± t (f; P) s _d --- (8) Here, t (f; P) is a value of a t-test used in statistics, f is a degree of freedom, and P is a risk factor. Represents f uses infinity ∞ or r-1 or N-1 described above. Generally, the risk factor P uses 5%, that is, 95% reliability. For example,
d = 0.04 min, s _d = 0.07 min, t (∞;
0.05) = 1.960, the "true difference δ is
0.04 ± 0.14 = [− 0.10, 0.18]. Is 95%.

Next, in step 75, the estimated section is interpreted as the certainty of the identification. If peak 5 and S
Even if there is a difference in the respective retention times of -A1c, the interval is at most -0.10 to 0.18 min. This means that peak 5 can be identified as S-A1c because there is 0.00 in the section. Further, if the peak 5 is not S-A1c, it means that there is a possibility that the difference in the true retention time is erroneously identified as an impurity separated by -0.10 to 0.18 min. In other words, 0.10 min on the negative side of the S-A1c peak.
This means that an impurity peak that is farther away or more than 0.18 min away from the positive side is not erroneously identified.

For convenience, when the value of this section is divided by the retention time of 1.34 min, “0.
07, there is a possibility that impurities were identified with a risk factor of 0.13 on the positive side. It can be said. Furthermore, it can be said that "no impurity is identified with a reliability of 0.93 on the negative side and 0.87 on the positive side of S-A1c".

Next, at step 76, the estimated section [-0.10, 0.18] or the reliability ratio [0.93, 0.87] defined for convenience is read from the CRT 46 or the printer 47.
Then, as shown in (A) or (B) of FIG.
In this way, the method of statistically examining peak identification ends (step 77).

Although the qualitative analysis has been described using the retention time and the error, the authenticity of the identification can be measured by referring to the peak area ratio. Fuzzy proposition "A1c is 3 ~
The area ratio is 15%. And so on to contribute to the identification probability. In addition to the data and credibility of the qualitative and quantitative analysis described above, the identification components and identification probabilities, and the quantitative values and quantitative errors, the reliability of the device or sample is also confirmed using fuzzy inference. be able to. For example, the proposition "Total area of all peaks is 100 to
3 million μV · s. And the like can be set to quantify reliability.

FIG. 14 shows the data analysis unit 45 shown in FIG.
3 is a functional block diagram of FIG. In FIG. 14, the storage unit 45
9 stores the data output from the detector 42. The peak identification unit 450 reads out the data stored in the storage unit 459 and identifies the peak of the chromatogram. The section calculation unit 451 includes a search section calculation unit 451A and a fit section calculation unit 451B. Then, the search section calculation unit 451 calculates a search section of the component L-A1c, for example. Also, the fit section calculation unit 451B
, A section in which the data point fits the cubic curve is calculated. The fitting unit 452 fits the data points to the cubic curve using the least squares method. Also,
The authenticity diagnosis unit 453 diagnoses the authenticity of the obtained numerical value.

The differential feature point search section 454 has a local maximum point search section 454A and an inflection point search section 454B. The coordinate calculating unit 455 calculates the coordinates of the local maximum point searched by the local maximum point searching unit 454A.
55A and an inflection point coordinate calculation unit 455B for calculating the coordinates of the inflection point searched by the inflection point search unit 454B.

The error / probability calculating section 456 includes a time error calculating section 456A, a fuzzy inference section 456B, a quantitative value error calculating section 456C for calculating an error of the quantitative value calculated by the quantitative value calculating section 457, and an identification probability. Calculation unit 456D
And Then, the error / probability calculation unit 456 calculates the peak holding time and the error corresponding to the identified maximum point and the calculated inflection point, the peak height and the error, the peak area and the error, the peak percentage and the error. Is calculated. The fuzzy inference unit 456B of the error / probability calculation unit 456 calculates the likelihood of the identified and detected local maximum point and inflection point using an index that is a probability, a difference, a confidence interval, or a confidence rate, and outputs the probability to the CRT 46 or the printer. Based on the index calculated in 47, a warning mark * or a double mark? Is displayed. Further, the data analysis unit 45 includes a baseline calculation unit 458 that calculates the above-described baseline. Then, the above-described analysis of the chromatogram is executed by the data analysis unit 45 having the above configuration.

The error / probability calculation unit 456 changes the error of the peak area according to the shape of the baseline for determining the peak area. Further, the error / probability calculator 456
Is input from the input unit 54 a statistical error to be included in the error calculation, a systematic error, and the like, and calculates the error of the peak holding time, the error of the peak height, the error of the peak area, and the percentage of the peak area percentage. The input statistical error and systematic error are added to the error to calculate these errors.

As described above, according to the embodiment of the present invention,
Chromatogram analysis display method and display that can determine quantitative values such as concentration and time such as retention time with high accuracy, find errors in the obtained values, calculate the reliability of quantitative values and peak identification, and display The device can be realized.

The following example may be used to calculate the error of the feature amount of the chromatogram.
That is, the number of theoretical plates is calculated in order to estimate the column efficiency. Also in this case, the error of each numerical value propagates, and an error of the theoretical plate number is generated. The formula for calculating the theoretical plate number is as follows (9
Equations -1) to (9-4) are known. ^{_{N = t R 2 / μ 2}} --- (9-1) N = (t R / σ) 2 --- (9-2) N = 5.54 × (t R / w 1/2) 2 - − (9-3) N = 16 × (t _R / w) ² −−− (9-4) where N is the number of theoretical plates, t _R is the retention time, μ ₂
Is the second moment, σ is half width at 60.7% of peak height, w _1/2 is full width at half maximum, w is width of tangent line from both inflection points to baseline and cut from the baseline It is.

To estimate the reliability of the number N of theoretical plates,
It is necessary to find the error. Also in this case, the error of each variable is first determined, and the error of the theoretical plate number is calculated.
The error in the retention time is required in any of the equations. Errors of other variables can also be determined statistically. In particular, in the case of Expression (9-4), it is desirable to use a multivariate analytical method using a standard error, a confidence limit, or the like in a case where a tangent from an inflection point is obtained. In addition, resolution Rs, capacity factor k ', selectivity α
And the like can be obtained in the same manner.

[0109]

Since the present invention is configured as described above, it has the following effects. Chromatogram analysis display method and display that can determine quantitative values such as concentration and time such as retention time with high precision, find errors in the obtained numerical values, calculate the reliability of quantitative values and peak identification, and display The device can be realized.

[Brief description of the drawings]

FIG. 1 is an operation flowchart of a chromatogram analysis and display method according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating a display example of a chromatogram analysis result on a CRT or a printer (plotter).

FIG. 3 is a schematic configuration diagram of an example of a glycated hemoglobin analyzer.

FIG. 4 is an explanatory diagram of a method of obtaining an error of a maximum point time.

FIG. 5 is a diagram illustrating another display example of a chromatogram analysis result on a CRT or a printer (plotter).

FIG. 6 is an explanatory diagram of a relationship between a baseline and an inflection point.

FIG. 7 is an explanatory diagram of an influence of a baseline on a quantitative value.

FIG. 8 is a diagram illustrating an example of a membership function.

FIG. 9 is a diagram showing another example of a membership function.

FIG. 10 is a diagram showing still another display example of calculated data.

FIG. 11 is an operation flowchart of a method for statistically testing peak identification.

FIG. 12 is a waveform chart illustrating an example of a second moment.

FIG. 13 is a diagram showing still another display example of calculated data.

FIG. 14 is a functional block diagram of a data analysis unit in the chromatogram analysis display device according to one embodiment of the present invention.

[Explanation of symbols]

 40 liquid sending pump 41 separation column 42 detector 43 sampler 44 control unit 45 data analysis unit 46 CRT 47 printer 51 unknown sample 52 standard sample 54 input unit 450 peak identification unit 451 section calculation unit 452 fitting unit 453 credibility diagnosis unit 454 differentiation Feature point searching unit 455 Coordinate calculating unit 456 Error / probability calculating unit 457 Quantitative value calculating unit 458 Baseline calculating unit 459 Storage unit

Continued on the front page (72) Inventor Hiroshi Satake 882 Ma, Katsuta-shi, Ibaraki Pref.Hitachi, Ltd.Measurement Instruments Division (72) Inventor Fujio Yamada 882 Katsuta-shi, Ibaraki Pref., Hitachi Measuring Instruments Division (72) Inventor Mitsuo Ito 882 Ma, Katsuta-shi, Ibaraki Pref.Hitachi, Ltd.Measurement Instruments Division (72) Inventor Shimane Watari Shimae 832-2 Hokuguchi, Horiguchi, Katsuta-shi, Ibaraki Pref. References JP-A-4-130270 (JP, A) JP-A-63-290958 (JP, A) JP-A-56-90259 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name ) G01N 30/86

Claims (32)

(57) [Claims] 1. A chromatogram analysis and display method for analyzing chromatogram data and displaying an analysis result, wherein: a step of detecting a peak of the chromatogram data; a step of identifying a peak of the chromatogram data; group peak size corresponding to peaks
Zui and, calculating an analysis value of the object to be measured, the error of the calculated analysis value of at least said peak
A chromatogram analysis display method, comprising: calculating using a size ; and displaying the calculated analysis value and an error of the analysis value on a display unit. 2. The chromatogram analysis and display method according to claim 1, wherein the error of the analysis value is calculated including the reliability of the baseline of the chromatogram data. 3. The chromatogram analysis and display method according to claim 1, wherein the error of the analysis value is calculated including the reliability of dividing peaks adjacent to each other. 4. The chromatogram analysis and display method according to claim 1, wherein said peak is an inflection point of the chromatogram data. 5. The chromatogram analysis and display method according to claim 1, further comprising a step of calculating the error of the peak size, and a step of displaying the calculated error of the peak size on a display unit. Chromatogram analysis display method. 6. The method according to claim 1 or 5 chromatogram analysis display method, wherein the peak size chromatogram analysis display wherein the peak is high. 7. The chromatogram analysis display method according to claim 1 or 5, wherein said peak size chromatogram analysis display method which is a peak area. 8. A chromatogram analysis display method according to claim 1 or 5, wherein said peak size chromatogram analysis display wherein the peak is high and the peak area. 9. A chromatogram analysis and display method for analyzing chromatogram data and displaying an analysis result, wherein: a step of detecting a peak of the chromatogram data; a step of identifying a peak of the chromatogram data; Calculating the calculated peak holding time; calculating the error of the calculated peak holding time; displaying the calculated peak holding time and the error of the holding time on a display unit. A method for analyzing and displaying a chromatogram, comprising: 10. The chromatogram analysis and display method according to claim 9, wherein the peak is an inflection point of the chromatogram data. 11. The method for analyzing and displaying chromatograms according to claim 9, wherein a step of calculating the reliability of peak identification based on the calculated peak retention time and an error of the peak retention time is provided. Displaying the reliability of the identification on the display means. 12. The chromatogram analyzing and displaying method according to claim 9, wherein the error of the calculated peak retention time is calculated including the reliability of dividing adjacent peaks. Analysis display method. 13. A chromatogram analysis and display device for analyzing chromatogram data and displaying an analysis result, comprising: a peak detection unit for detecting a peak of the chromatogram data; and a peak identification unit for identifying a peak of the chromatogram data. , based on peak size corresponding to the detected and identified peaks
Then , a quantitative value calculation unit that calculates the analysis value of the measured object, at least using the peak size , the calculated
An error calculation unit that calculates an error analysis, chromatogram analysis display apparatus comprising: an analysis value which is the calculated, and a display unit for displaying the error in the analysis. 14. The chromatogram analysis and display device according to claim 13, wherein the error of the analysis value is calculated including the reliability of the baseline of the chromatogram data. 15. The chromatogram analysis and display device according to claim 13, wherein the error of the analysis value is calculated including the reliability of dividing adjacent peaks. 16. The chromatogram analysis display device according to claim 13 , wherein said peak is an inflection point of the chromatogram data. 17. The chromatogram analysis and display device according to claim 13 , wherein said error calculation section calculates an error of the peak size, and said display section further displays the calculated error of the peak size. Characteristic chromatogram analysis display device. 18. The chromatogram analysis display apparatus according to claim 13 or 17, wherein said peak size chromatogram analysis display apparatus, wherein the peak is high. In the chromatogram analyzing display device 19. The method of claim 13 or 17, wherein said peak size chromatogram analysis display apparatus, characterized in that the peak area. 20. A chromatogram analysis display apparatus according to claim 13 or 17, wherein said peak size chromatogram analysis display apparatus, wherein the peak is high and the peak area. 21. A chromatogram analysis and display device for analyzing chromatogram data and displaying an analysis result, comprising: a peak detection unit for detecting a peak of the chromatogram data; and a peak identification unit for identifying a peak of the chromatogram data. A retention time calculation unit that calculates a retention time of the detected and identified peak; an error calculation unit that calculates an error of the calculated peak retention time; a retention time of the calculated peak; and the retention time And a display unit for displaying an error of the chromatogram. 22. A chromatogram analysis and display device according to claim 21 , wherein said peak is an inflection point of the chromatogram data. 23. The chromatogram analysis and display device according to claim 21, wherein the error calculating section calculates the reliability of peak identification based on the calculated peak holding time and the error of the peak holding time. The display unit further displays the calculated reliability of peak identification. 24. The chromatogram analyzing and displaying apparatus according to claim 21, wherein the error of the calculated peak retention time is calculated including the reliability of dividing adjacent peaks. Analysis display device. 25. A chromatogram analysis and display method for analyzing chromatogram data and displaying an analysis result, comprising: a step of identifying a maximum point of the chromatogram data within a predetermined range; and a step near the identified maximum point. Calculating a search section that does not include the maximum point; calculating a polynomial curve that fits the chromatogram data in the calculated search section; detecting an inflection point of the polynomial curve; And the peak holding time corresponding to the detected maximum point and inflection point, the error of the peak holding time, the peak height,
A peak calculation step for calculating a peak area, a peak area percentage, and an error of the peak area percentage; an error of the calculated peak retention time, a peak height, a peak area, a peak area percentage, and a peak area percentage. And c. Displaying on a display means a chromatogram analysis and display method. 26. The chromatogram analysis display method according to claim 25 , wherein said peak calculation step calculates a peak height error and a peak area error, and said display step includes the peak holding time, the peak holding time A chromatograph characterized by displaying on a display means an error of a retention time, an error of a peak height, an error of a peak height, a peak area, an error of a peak area, a peak area percentage, and an error of a peak area percentage. Gram analysis display method. 27. The method according to claim 25, wherein the step of calculating the search section comprises:
Determining a polynomial curve between the step of determining whether there is a local maximum point in the calculated search section,
Setting a search section obtained by extending the search section by a predetermined width when there is no local maximum point in the search section. 28. A method according to claim 25, wherein the likelihood of the identified and detected maximum point and inflection point is calculated by an index which is a probability, a difference, a confidence interval or a confidence rate. Displaying the calculated index on a display means. 29. A chromatogram analysis and display device for analyzing chromatogram data and displaying the analysis result, comprising: a detecting means for detecting a component contained in the sample; and a storage unit for storing the chromatogram data detected by the detecting means. A peak identification unit that identifies a local maximum of data within a predetermined range of the chromatogram data stored in the storage unit, and calculates a search section that is near the identified local maximum and does not include the local maximum. An interval calculating unit; a fitting unit that calculates a polynomial curve that fits the chromatogram data in the calculated search interval; a differential feature that calculates a differential value of the polynomial curve and detects an inflection point from the calculated differential value A point search unit, a peak holding time corresponding to the identified maximum point and the inflection point, an error in the peak holding time, a peak height, and a peak plane. An error probability calculation unit that calculates the product, the peak area percentage, and the error of the peak area percentage; and the calculated peak holding time, peak height, peak area, peak area percentage, and peak area percentage error. And a display unit for displaying: and a chromatogram analysis display device. 30. The chromatogram analysis and display device according to claim 29 , wherein said error probability calculating section calculates an error of a peak height and an error of a peak area, and said display section calculates said peak holding time, Chromatogram analysis characterized by displaying a peak retention time error, a peak height, a peak height error, a peak area, a peak area error, a peak area percentage, and a peak area percentage error. Display device. 31. The chromatogram analysis and display device according to claim 29, wherein the error probability calculation unit determines the likelihood of the identified and detected maximum point and inflection point by a probability, a difference, a confidence interval or a confidence rate. Calculated by a certain index, the display unit
A chromatogram analysis and display device for displaying the calculated index. 32. The chromatogram analysis display device according to claim 29 , wherein said error probability calculation section changes the error of the peak area according to the shape of the baseline for determining the peak area. Display device.